Universidade de So Paulo

Biblioteca Digital da Produo Intelectual - BDPI

Departamento de Fsica e Cincia Interdisciplinar - IFSC/FCI

Artigos e Materiais de Revistas Cientficas - IFSC/FCI

2012-02

Characterization of the glass-to-vitroceramic

transition in yttrium aluminum borate laser
glasses using solid state NMR
Solid State Nuclear Magnetic Resonance,London : Academic Press,v. 41, p. 48-59, Feb. 2012
http://www.producao.usp.br/handle/BDPI/49504
Downloaded from: Biblioteca Digital da Produo Intelectual - BDPI, Universidade de So Paulo

Solid State Nuclear Magnetic Resonance 41 (2012) 4859

Contents lists available at SciVerse ScienceDirect

Solid State Nuclear Magnetic Resonance

journal homepage: www.elsevier.com/locate/ssnmr

Characterization of the glass-to-vitroceramic transition in yttrium aluminum

borate laser glasses using solid state NMR
Heinz Deters a,b, Hellmut Eckert a,n
a
b

Institut f
ur Physikalische Chemie
Graduate School of Chemistry, Westf
alische Wilhelms-Universit
at M
unster, Corrensstrasse 30, D48149 M
unster, Germany

a r t i c l e i n f o

a b s t r a c t

Article history:
Received 11 August 2011
Received in revised form
14 November 2011
Available online 9 December 2011

The crystallization of laser glasses in the system (Y2O3)0.2{((Al2O3)x)(B2O3)0.8  x} (0.15 r x r0.40) doped
with 0.5 mol% of ytterbium oxide has been investigated by x-ray powder diffraction, and various solid
state NMR techniques. The crystallization process has been analyzed in a quantitative fashion by highresolution solid state 11B, 27Al, and 89Y NMR spectroscopy as well as 11B{27Al} and 27Al{11B} rotational
echo double resonance (REDOR) experiments. The homogeneous glasses undergo major phase
segregation processes resulting in crystalline Al5BO9 (historically denoted as Al18B4O33), YBO3, crystalline YAl3(BO3)4, residual glassy B2O3, and an additional yet not identied crystalline phase (X-phase).
& 2011 Elsevier Inc. All rights reserved.

1. Introduction
Rare earth (RE) doped yttrium borate (YB) and yttrium
aluminoborate (YAB) crystals are of great interest because of
their emission properties in combination with non-linear optic
effects [13]. As the utility of single crystalline materials is often
compromised by comparatively high production costs and the
limited solubility of the RE ions in the single crystalline hosts,
research interest has shifted towards the preparation of glassy
and ceramic materials, which can often accommodate larger
concentrations of the RE dopants [4,5]. Following this idea, glasses
of the ternary system Y2O3Al2O3B2O3 have been introduced as
excellent alternatives to single crystalline materials for special
laser applications involving self-frequency doubling or self-sum
frequency mixing [6]. By high-temperature annealing these
glasses can be converted to vitroceramics, which offer the
distinctive advantage of presenting higher emission cross-sections owing to the more ordered local crystalline environment.
While the preparation conditions, the glass forming range, as well
as some basic bulk and spectroscopic properties of Y2O3Al2O3
B2O3 glasses have been published [711] no corresponding
structural description of the vitroceramics derived from them is
as yet available. Owing to its element-selectivity, inherently
quantitative character and its focus on the local environment,
solid state nuclear magnetic resonance (NMR) is an ideal tool
for structural investigation of vitroceramics, and numerous applications of single- and double resonance applications to glass

Corresponding author.
E-mail address: eckerth@uni-muenster.de (H. Eckert).

0926-2040/$ - see front matter & 2011 Elsevier Inc. All rights reserved.
doi:10.1016/j.ssnmr.2011.11.008

ceramics can be found in the literature [1217]. In contrast to

diffraction techniques, solid state NMR methods are equally
sensitive to the glassy and crystalline constituents, resulting in
quantitatively representative spectra. Benetting from the potential of advanced double resonance solid state NMR strategies to
provide useful structural information on a local level [18,19], we
have recently developed a detailed description of the framework
structure of yttrium aluminoborate glasses [20]. While 11B and
27
Al magic-angle spinning (MAS) NMR experiments are informative regarding the coordination states of the boron and aluminum
atoms, complementary 11B/27Al double resonance NMR studies
have given quantitative information regarding the boron
aluminum connectivities in the network. In the present study,
the structure and quantitative phase composition of yttrium
aluminum borate vitroceramics will be discussed in relation to
such NMR data, leading to a quantitative description of the
crystallization product distribution.

2. Experimental
2.1. Sample preparation and characterization
Glasses along the composition line (Y2O3)0.2{(Al2O3)x(B2
O3)0.8  x} (0.15rx r0.40), containing Yb at variable dopant levels
were prepared as described previously [6,20], by the conventional
melt-quenching technique, using a resistive furnace open to the
atmosphere. To obtain a water free starting material, glassy boron
oxide was prepared by annealing commercial B2O3 (ALDRICH,
99%) at 450 1C for 2 h, with a heating rate of 5 1C/min, for the
evaporation of water. Next, it was melted at 900 1C, subsequently

H. Deters, H. Eckert / Solid State Nuclear Magnetic Resonance 41 (2012) 4859

Table 1
Sample compositions, glass transition (Tg) and crystallization temperatures (Tx) of
the yttrium aluminoborate glasses under study.
Sample

Y2O3/mol%

Al2O3/mol%

B2O3/mol%

Tg (7 5)/K

Tx (7 5)/K

Al15
Al20
Al25
Al30
Al35
Al40

20
20
20
20
20
20

15
20
25
30
35
40

65
60
55
50
45
40

983
985
992
1004
1016
1028

1197
1169
1187
1219
1201
1208

quenched on a copper plate, and stored in an oven at 200 1C.

Glassy boron oxide, Al2O3 (FLUKA, 99 %), and Y2O3 (ACROS
ORGANICS, 99.99%) were dry mixed and subsequently melted in
a covered platinum crucible at 15001600 1C for 3040 min.
Subsequently, the melts were rapidly quenched between two
copper plates. The weight loss due to evaporation did not exceed
2% for any of the samples. Glass transition and crystallization
temperatures were determined by differential thermal analysis
(DTA) using a Netzsch STA 409 C/CD instrument operated at a
heating rate of 10 1C/min. The homogeneous glasses were
annealed at 1150 1C for 60 min and 1440 min for the pure and
the Yb doped glasses, respectively, resulting in the vitroceramic
(VC) samples (see Table 1). The model compound Al5BO9 (historically known as Al18B4O33, see discussion below) was prepared as
described in the literature for Al18B4O33. Stoichiometric amounts
of glassy (water free) B2O3 and Al2O3 (FLUKA, 99 %) were ground
to obtain a homogeneous mixture, which was pressed and
annealed at 1150 1C for 10 hours thereafter. X-ray diffractograms,
using an ENRAF NONIUS FR552 diffractometer, operated with a
CuKa source and a GUINIER camera, did not reveal any secondary
phases (PEARSON Crystal Database, no. 1401325) [21].

2.2. Solid state NMR

11
B and 27Al solid state NMR spectra were acquired at 160.5
and 130.3 MHz, respectively, on a BRUKER AVANCE DSX 500 FTNMR spectrometer (11.74T) equipped with various 4 mm MASNMR probes operated at a typical spinning frequency of 15 kHz.
Short excitation pulses (corresponding to solid ip angles below
301) were used. 11B triple-quantum (TQ)-MAS spectra were
obtained using the three-pulse z-ltering sequence [2224] at a
MAS rotation frequency of 25 kHz. Triple quantum excitation and
re-conversion to zero-quantum coherence was effected at a
nutation frequency of 192 kHz (for a liquid sample) with pulses
of 2.1 ms and 0.8 ms length, respectively, and the signal detected
with a soft pulse of 10 ms length. The evolution time was
incremented in steps of 10 ms, and 480 data sets were acquired
in the t1 dimension, using 24 scans at a recycle delay of 10 s. 27Al
triple-quantum (TQ)-MAS experiments were also carried out
using the z-ltering sequence, at a MAS rotation frequency of
1415 kHz. In this case, the nutation frequency of 27Al was
120 kHz (for the liquid reference sample), and the lengths of the
rst two pulses were 3.0 and 1.1 ms, respectively. The single
quantum signal was detected by a soft pulse of 10 ms length. For
these measurements the evolution time was incremented in steps
of 8.3316.67 ms, and 100300 data sets were acquired in the t1
dimension, using 72540 scans (depending on the Al content in
the sample) and a recycle delay of 2.0 s. All the line shape
analyses were performed using the DMFIT program [25]. Following data shearing, the resonance shifts in the isotropic (F1) and in
the anisotropic (F2) dimensions of the 2D plots are given by:

dF1 34diso
CS 60dQ =9

1a

dF2 diso
CS 3dQ

49

1b

where

dQ 3SOQE2 =10n2L 2I2I12

1c

following the description by Medek and Frydman [26]. In these

expressions the second-order quadrupolar effect SOQE is dened
as CQ(1 Z2/3)1/2. Chemical shifts diso
CS and the quadrupolar coupling constants CQ and the asymmetry parameters Z for individual
11
B resonances were obtained by tting the lineshapes observed
in the F2 dimension for each resolved 1D slice dened by the
frequency for which the intensity maximum is observed in the F1
dimension. 11B and 27Al isotropic chemical shifts are externally
referenced relative to BF3 Et2O and 1 M aqueous Al(NO3)3 solutions, respectively.
11
B{27Al} and 27Al{11B} rotational echo double resonance
(REDOR) experiments were conducted with the BRUKER AVANCE
DSX 500 FT-NMR spectrometer (11.74T) at a spinning frequency
of 14 kHz. In both cases the conventional REDOR sequence of
GULLION and SCHAEFER [27] was used. In the 11B{27Al} REDOR
experiments the 1801 pulse lengths for the 11B observe nuclei
were 5.56.2 and 7.27.8 ms (nutation frequency for the liquid
standard: 50 kHz) for the three- and four-coordinated boron
species, respectively, and a recycle delay of 40 s was used. The
27
Al 1801 pulse lengths were 5.86.5 ms (nutation frequency for
the liquid standard: 28 kHz) and the pulses were phase cycled
according to the XY-4 scheme [28]. 27Al{11B} REDOR experiments
were done with 1801 pulse lengths of 5.86.5 ms and 5.56.2 ms
for 27Al and 11B, respectively, and a recycle delay of 1.0 s was
used. The pulses on the non-observe 11B channel were also phase
cycled according the XY-4 scheme.
89
Y MAS-NMR spectra were acquired at 24.5 MHz on a
BRUKER AVANCE DSX 500 FT-NMR spectrometer (11.74T), using
a 7 mm probe specially designed for low-gamma nuclei, which
was operated at spinning frequencies of 57 kHz. Since zirconia
rotors contain small yttrium contents (23%) [29] leading to a
strong background signal, Si3N4 rotors were used for all 89Y NMR
experiments. In the case of ceramic samples, 8 scans were
accumulated as rotor-synchronized spin echoes using 901 (1801)
pulses of 9 (18) ms length and a relaxation delay of 5000 s. For the
RE doped glassy samples 8400 scans were accumulated with a
relaxation delay of 10 s. Chemical shifts are referenced against
1 M YCl3 aqueous solution, using an 8 M Y(NO3)3 aqueous solution containing 0.25 M Fe(NO3)3 for relaxation enhancement as a
secondary standard. (d  22.8 ppm).

3. Results and discussion

3.1. Model compound studies: Al5BO9
3.1.1. 11B and 27Al MAS and TQMAS NMR
For Al5BO9 (historically known as Al18B4O33 9Al2O3  2B2O3),
one crystallographically unique orthoborate BO33  group is present in the crystal structure [21,30]. This trigonal boron site gives
rise to a broad 11B resonance whose line shape represents a
second-order quadrupolar powder pattern, as shown in the left
part of Fig. 1. The local symmetry of this orthoborate group
results in a nearly axially symmetric electric eld gradient. The
isotropic chemical shift and quadrupolar coupling parameters can
directly be obtained by tting the MAS NMR spectrum. We obtain
diso
CS 16.5 ppm, CQ 2.60 MHz and Z 0.09. In addition, a very
small extra resonance near 0 ppm can be identied, indicating
the presence of a small amount (o2%) of fourfold-coordinated
boron. Table 2 summarizes the 11B solid state NMR parameters

50

H. Deters, H. Eckert / Solid State Nuclear Magnetic Resonance 41 (2012) 4859

0
5

15
20

/ppm

10

25

30

20

10

35

-10

35 30 25 20 15 10 5
F2
/ppm

(11B) / ppm
Fig. 1. Left:

11

B MAS NMR spectrum of Al5BO9 with a corresponding simulation (red curve). Right:

11

F1

30
Al5BO9

0 -5 -10 -15

B TQ MAS NMR spectrum of Al5BO9.

Table 2
11
11
B isotropic chemical shifts diso
B MAS
CS , quadrupolar coupling constants CQ, and electric eld gradient asymmetry parameters Z obtained from lineshape analysis of the
NMR spectra of crystalline model compounds. In addition the tting parameters G/L (fraction of Gaussian (vs. Lorentzian) character), and LB (line broadening
parameter full width at half-maximum) of the convolution function used.
Sample

Group

diso
CS (7 0.2)/ppm

G/L

LB (7 10)/Hz

CQ (7 0.1)/MHz

g (70.05)

11

YBO3
YAl3(BO3)4

B(4)
B(3)-I
B(3)-II
B(3)
B(3)
B(3)-I
B(3)-II

0.4
20.1
18.9
16.5
19.5
18.1
14.0

0.32
-

300
28
95
23
137
208
160

3.0
2.9
2.6
3.0
2.7
2.6

0.08
0.07
0.09
0.09
0.20
0.22

100
27
73
100
100
70
30

Al18B4O33
X-phase
B2O3 glass

B NMR area/%

-60
-40

Al(6)

-20
0
40

Al(5)-I
Al(5)-II

/ppm

20
60
80

Al(4)

120
120100 80 60 40 20 0 -20 -40 -60
F2

F1

100
100

50

-50

(27Al) / ppm

/ppm

Fig. 2. 27Al TQ MAS NMR spectrum obtained for crystalline Al5BO9 (left). Line shapes observed in the F2 dimension for each resolved F1 slice (right). The corresponding
simulations are included. In case of vefold coordinated aluminum sites, the resonance has to be tted by two components as displayed by two dotted curves. Spinning
sidebands are marked by asterisks.

for all the crystalline borate reference compounds relevant to the

present study.
Regarding the local environment of the aluminum species, the
crystal structure of Al5BO9 indicates the presence of one four-, one
six-, and two ve fold coordinated aluminum sites in the asymmetric unit. Each of these aluminum sites gives rise to a second-

order quadrupolar powder pattern resulting in a complicated

MAS NMR spectrum shown in Figs. 2 and 3. A signicantly
enhanced resolution can be achieved using the 27Al TQ MAS
NMR experiment. In a rst approximation, the TQ MAS NMR
spectrum illustrated in Fig. 2 reveals only three different aluminum resonances that are spectroscopically resolved in the

H. Deters, H. Eckert / Solid State Nuclear Magnetic Resonance 41 (2012) 4859

isotropic F1 dimension. Upon closer examination, however, a

strong overlap of two resonances in the F1 dimension can be
noticed by tting the individual cross sections of the 2D-plot
along the F2 dimension, as shown in the right part of Fig. 2. These
resonances can be attributed to the two vefold coordinated
aluminum sites. All isotropic chemical shifts and quadrupolar
coupling parameters obtained by evaluating the TQ MAS NMR
spectrum in the previously described manner are listed in Table 3,
in good agreement with literature data, regarding previous 27Al
NMR measurements on samples denoted as Al18B4O33 [3135].
Based on these parameters the rather complicated MAS NMR
spectrum can be simulated (Fig. 3), thereby yielding the quantitative deconvolution of the individual sites. The results are also
summarized in Table 3 and are in very good agreement with the
crystal structure [21,30].
The foregoing data are important in connection to a current
on-going discussion regarding a discrepancy between the crystal
structure and the exact chemical composition of this compound,
suggesting the presence of some structural disorder [3638].
Chemical analyses have repeatedly suggested a composition of
Al18B4O33 (Al4.91B1.09O9), whereas the structural data derived
from single-crystal X-ray diffraction studies (mullite structure)
are consistent with the nominal formula Al5BO9 [21]. This inconsistency has been explained by a statistical replacement of some
aluminum atoms by boron atoms [21]. Recently it has been
attempted to identify the different boron environments arising
from the B/Al replacement in the decient crystal structure
[36,37]. Using solid state NMR You et al. [36] claimed the

51

presence of an additional B(4) site in this compound; however,

an inspection of his data gives no evidence for this claim. While
our solid state NMR results indeed show a small extra
B(4) resonance near 3 ppm, the fractional area of this site is less
than 2%, i.e. clearly inconsistent with the model of You et al.
(according to which 9% if the boron would be four-coordinate,
substituting Al(4) by B(4) according to the compositional formula
Al4.91B1.09O9). Recently the question about the correct chemical
stoichiometry of this compound has been addressed comprehensively by Fisch et al. [38], on the basis of combined single-crystal
diffraction and chemical analysis data. The authors concluded
that the correct stoichiometry is somehow in between Al18B4O33
and Al5BO9 and they attributed the deviation from Al5BO9
composition to a very low amount of Al vacancies at a ve-fold
coordinated site, which is charge balanced by an additional
nearby three-fold coordinated B site [38]. Using solid state
NMR, the authors were unable to detect B(4) units. This prediction is in better agreement with the results from our study
assuming that the extra trigonal boron sites are not separately
detectable in the 11B NMR experiments. Unlike Fisch et al. and
others, we do detect a small amount of B(4) units, although the
quantication of this spectral component leads to the conclusion
that the extent to which boron is substituting on the Al(4) sites
must be less than 2%. Our 27Al MAS-NMR analyses are in general
agreement with previous work done on this compound; however,
the NMR quantication of these sites by 27Al MAS NMR is neither
sufciently accurate nor sufciently precise to elucidate the exact
relative occupancies of the various aluminum sites present.
3.1.2. 11B{27Al} and 27Al{11B}REDOR experiments
The 11B{27Al} REDOR results obtained for Al5BO9 and
YAl3(BO3)4 measured for comparison are illustrated in Fig. 4. As
previously discussed, REDOR data in multispin systems can be
analyzed with the parabolic approximation [39] applied to the
short-term evolution behavior (0 r DS/So r0.2) using the expression [18,19,40]:

DS

1
NT R 2 Mi2hetero
S0 II 1p2

Al5BO9

100

50

-50

(27Al) / ppm
Fig. 3. 27Al MAS NMR spectrum of crystalline Al5BO9. Red curve illustrates a
simulation based on four components, which are individually displayed as dotted
curves (For interpretation of the references to colour in this gure legend, the
reader is referred to the web version of this article.).

where I denotes the spin quantum number of the non-observed

nucleus. The theoretical second moments M2 can be calculated
from the known crystal structure using the van Vleck equation:

2
X
4 m0 gi gj _
Mi2hetero
II 1 r 6
3
ij
4p
15
ij
For quadrupolar nuclei, the analysis is more complicated;
however, as owing to the anisotropic quadrupolar interactions
the p pulses applied to the non-observed I spins that are present
in the non-central Zeeman states may provide less dipolar
recoupling than theoretically expected or even none at all,
requiring careful calibration procedures on crystalline model
compounds with well-characterized spin dynamics [18,19,40].
To this end, the theoretical second moments and those obtained

Table 3
27
Al MAS NMR parameters of Al5BO9 and YAl3(BO3)4. LB represents the line broadening parameter of the convolution function used.
Sample

Group

diso
CS (7 1.0)/ppm

LB (7 10)/Hz

CQ (7 0.3)/MHza

ga (7 0.05)

27

Al5BO9

Al(4)
Al(5)-I
Al(5)-II
Al(6)
Al(6)

70.8
53.3
49.0
9.0
1.2

331
197
261
267
702

8.9
6.9
8.4
5.8
1.8

0.44(0.50)
0.07 (0.03)
0.45 (0.73)
0.51 (0.53)
not det.

20 (20)
18 (20)
23 (20)
39 (40)
100

YAl3(BO3)4
a
b

Values in brackets are calculated using the WIEN2k code.

Values in brackets are the expected area fractions from the respective crystal structure.

(8.6)
(7.8)
(7.6)
(5.7)

Al NMR area (7 5%)/%b

52

H. Deters, H. Eckert / Solid State Nuclear Magnetic Resonance 41 (2012) 4859

1.0

0.8

Al5BO9

Al5BO9

0.8
S/S0

S/S0

0.6
0.4
YAl3(BO3)4
0.2

YAl3(BO3)4

0.6
0.4
0.2

0.0

0.0

0.000

0.001

0.002
0.003
NTR / s

0.004

0.005

0.00

0.01

0.03

0.02
NTR / s

Fig. 4. 11B{27Al} REDOR curves of the model compounds YAl3(BO3)4 and Al5BO9. Left: The solid curves represent parabolic ts to Eq. (1) in the limit of short dipolar
evolution times. Right: REDOR attenuations of the two crystalline model compounds are found to be complete at longer dipolar evolution times. All REDOR measurements
were done at a rotation frequency of 14 kHz.

Table 4
11
B {27Al} REDOR results on crystalline model compounds and theoretical second
moments. Experimental second moments M2 and calibration factors f were
determined in the limit of short dipolar evolution times (0 r DS/S0 r0.1).
a
Weighted average for the two crystalline boron sites.
6
theor/10

Sample

Group

M2,

YAl3(BO3)4

B(3)-I
B(3)-II
B(3)

83.0
64.6
101.7

Al5BO9

Table 5
27
Al {11B} REDOR results of crystalline model compounds and theoretical M2
values. Experimental second moments M2 and calibration factors f were determined in the limit of short dipolar evolution times (0 r DS/S0 r 0.2) by parabolic
ts according to Eq.(2).

rad2s  2

M2, exp/106 rad2s  2

(7 10%)

Sample

Group

M2,

69.2a

7.6

0.110

YAl3(BO3)4
Al5BO9

Al(6)
Al(6)
Al(5)-I
Al(5)-II
Al(4)

39.5
12.1
2.61
14.1
2.52

11.0

0.108

YAl3(BO3)4

theor/10

rad2s  2

6.4

M2, exp/106 rad2s  2

( 710%)

7.9
1.8
0.9

0.201
0.145
0.141

Weighted average for the three aluminum sites (Al(5)-I, Al(5)-II, and Al(4)).

0.6
Al5BO9:
0.4
S/S0

Al(6)

0.2

Al(5)-I, Al(5)-II, and Al(4)

0.0
0.000

0.001

0.002

0.003

NTR / s
Fig. 5. 27Al{11B} REDOR curves obtained for crystalline YAl3(BO3)4 and Al5BO9
(2 curves, see text). The solid curves represent parabolic ts to Eq. (2) in the limit
of short evolution times.

experimentally are listed in Table 4 together with the empirical

calibration factors f representing the ratio between the experimental and the calculated M2 values. It is somewhat surprising
that the calibration factors are more or less identical for the two
measured crystalline model compounds, since the involved quadrupolar coupling parameters differ signicantly. While the quadrupolar coupling constant for the Al(6) site in YAl3(BO3)4 is rather
small, strong quadrupolar couplings are present for all aluminum
sites in Al5BO9 (see Table 3).
Fig. 5 summarizes the 27Al{11B} REDOR results obtained for
crystalline YAl3(BO3)4 and Al5BO9. In the case of Al5BO9, four
different aluminum resonances can principally be distinguished

in the 27Al MAS NMR spectrum, as shown in Fig. 4, but a strong

overlap of the Al(5)-I, Al(5)-II, and Al(4) resonances results in a
low spectroscopic resolution. In REDOR experiments, intensity
differences of individual resonances are measured and hence
measurement at higher magnetic eld strengths will be necessary
in order to get high-accuracy results. In the present study we have
addressed this problem by extracting an average M2 value from
the data measured for the three aluminum sites for which the 27Al
MAS-NMR lineshapes are overlapping, and comparing it with the
average calculated value from the crystal structure considering
the 1:1:1 ratio of the site populations. On the other hand, the
Al(6) signal is spectroscopically well resolved and hence the
separate REDOR attenuation curve of this particular resonance
can be analyzed individually.
Table 5 summarizes the theoretical and experimental second
moments as well as the resulting calibration factors determined by
comparing the experimental and calculated values with each other.
While the calibration factors obtained for the two REDOR curves
associated with crystalline Al5BO9 are in very good agreement with
each other, they clearly deviate from the factor obtained for
crystalline YAl3(BO3)4. This deviation is probably related to the
different magnitudes of the dipolar coupling constants involved.
Furthermore, the initial REDOR curve of YAl3(BO3)4 reveals substantial deviations from parabolic behavior, and thus the tting
results can only be considered as a qualitative suggestion.
3.2. Vitroceramics
3.2.1. XRD results
The XRD data of the yttrium aluminoborate vitroceramics in
the VC-Alx series are summarized in Fig. 6. The data reveal the

H. Deters, H. Eckert / Solid State Nuclear Magnetic Resonance 41 (2012) 4859

crystallization of both YBO3 and YAl3(BO3)4 for all of the vitroceramic samples studied. In the case of vitroceramics with high
alumina contents (VC-Al40), weak additional diffraction peaks
can be observed, which can be assigned to the formation of
crystalline Al5BO9. Note that the signal-to-noise ratio of these
reections is very low due to the weak scattering factors of
aluminum, boron, and oxygen atoms for X-rays. However, no
other diffraction peaks are seen in the X-ray diffractograms,
suggesting that no additional crystalline phases are formed at
levels above XRD detection limits.
3.2.2. 11B MAS and TQMAS NMR Spectra and 11B{27Al} REDOR
Fig. 7 (left) shows the TQ MAS NMR spectrum of glass Al40
before and after crystallization. In the glassy state, the dominant
fraction of boron is three-coordinated, giving rise to a broad
structured lineshape characterized by second-order quadrupolar
perturbations. As previously discussed, the lineshape suggests a
non-axially symmetric electric eld gradient, consistent with the
formation of mostly metaborate (three-coordinate boron with one
non-bridging oxygen) and pyroborate (three-coordinate boron
with two non-bridging oxygen) groups [20]. The chemical shift
difference between the latter two units appears to be obscured by
a wide distribution of quadrupolar coupling parameters and
chemical shifts. In addition, only a small fraction of four-coordinated boron sites can be identied.

YBO3
YAl3(BO3)4
Al5BO9
VC-Al40
VC-Al35
VC-Al30
VC-Al25
VC-Al20
VC-Al15
20

30

40
2 /

50

60

Fig. 6. Powder XRD data of the yttrium aluminoborate vitroceramics in the VC-Alx
series and the crystalline model compounds Al5BO9, YAl3(BO3)4, and YBO3.

53

The glass-to-vitroceramic transition results in a marked

change of the 11B NMR spectrum, producing several threecoordinated boron species with close to axially symmetric electric
eld gradients, in addition to four-coordinated boron. Fig. 7
shows the spectra for the entire vitroceramic series, indicating
systematic changes in the boron coordination environment with
varying composition. Additional spectroscopic resolution is provided by the 11B TQ MAS NMR results (Fig. 8), which basically
reveal the presence of crystalline YBO3 (peak at diso
CS 0.4 ppm and
YAl3(BO3)4 (two peaks in a 3:1 ratio at diso
CS 18.9 and 20.1 ppm))
throughout all of the samples investigated. In case of ceramics
with high alumina contents (0.3 rxr0.4) Al5BO9 is unambiguously detected as an additional phase, whereas two broader
resonances at 14.0 and 18.1 ppm are detected in samples with
xo0.30. As previously discussed for other vitroceramics in the
Y2O3B2O3Al2O3 ternary system, these resonances are identical
to those found in glassy B2O3 and can be easily attributed to the
B(3) units inside and outside of the boroxol rings, respectively
[41]. Closer examination reveals an additional sharp signal in the
F1 dimension of the spectra of all the samples investigated. The
chemical shift and quadrupolar coupling parameters of this
particular resonance can be obtained by tting the respective
lineshape observed in the corresponding F2 dimension. In doing
so, the isotropic chemical shift is found to be located at about
19.5 ppm; CQ is about 3.0 MHz and the asymmetry parameter Z is
about 0.09, suggesting a close to axially symmetric electric eld
gradient. Obviously, this resonance is attributable to a single
three-coordinated boron site that is present in the whole set of
vitroceramic samples. The presence of crystalline Al4B2O9 can be
ruled out as this would result in an additional resonance for the
four-coordinated boron species [42]. Another possibility is AlBO3,
the structure of which is based on a single orthoborate BO33  site
with D3h local symmetry [43]. While this is nicely consistent with
the NMR parameters extracted from Fig. 8, no reference NMR data
(isotropic chemical shift, CQ value) on this crystalline compound
are available at the present time. Thus, an assignment of this
resonance to this compound would be rather speculative. Finally,
the additional 11B resonance might also arise from Y8.67B2
(BO3)2O13, in which four crystallographically distinct B(3) units
are present [44] but on which no NMR data are available either.
Since a denite assignment cannot be done at the present stage,
this resonance will be attributed to an unknown phase X, whose
identication will have to be left for future investigations.
The quantitative deconvolution of the 11B MAS NMR spectra of
the vitroceramics is possible using the tting parameters
obtained for crystalline YBO3, YAl3(BO3)4, Al5BO9, glassy B2O3
and for the unknown phase X, leaving only the amplitudes of the

Al5BO9
VC-Al40
VC-Al35
Al40 (glass)
VC-Al30
VC-Al25
VC-Al20
VC-Al40
VC-Al15
60

40

20

-20

-40 60

40

(11B) / ppm
Fig. 7. Left: glass-to-vitroceramic transition observed for the sample with x 0.4 (40 mol% of Al2O3) by
aluminoborate vitroceramics in the VC-Alx series and of crystalline Al5BO9.

20

-20

-40

(11B) / ppm
11

B MAS NMR. Right:

11

B MAS NMR spectra of yttrium

H. Deters, H. Eckert / Solid State Nuclear Magnetic Resonance 41 (2012) 4859

0
YBO3

20

YAl3(BO3)4

25

Phase X

30

35

35
35 30 25 20 15 10 5 0 -5 -10 -15

/ppm

/ppm

F2

0
YBO3

20

20

25

Phase X 25

30

30

35

35
35 30 25 20 15 10 5 0 -5 -10 -15

/ppm

/ppm

F2

5
10

20
Phase X

35 30 25 20 15 10 5 0 -5 -10 -15
F2

/ppm

5
10

glassyB2O3
15
YAl3(BO3)4

20
Phase X

25

25

30

30

35

35

F1

YAl3(BO3)4

/ppm

glassyB2O3
15

YBO3

enlarged

enlarged

YBO3

/ppm

15

YAl3(BO3)4

35 30 25 20 15 10 5 0 -5 -10 -15
F2

10
glassyB2O3

F1

Phase X

YAl3(BO3)4

F1

15

/ppm

10
Al5BO9

enlarged

enlarged

YBO3

25

30

35 30 25 20 15 10 5 0 -5 -10 -15
F2

15

/ppm

Phase X

Al5BO9

/ppm

20

YAl3(BO3)4

10

F1

15

F1

Al5BO9

/ppm

10

enlarged

enlarged

YBO3

F1

54

35 30 25 20 15 10 5 0 -5 -10 -15
F2

/ppm

Fig. 8. 11B TQ MAS spectra obtained for the vitroceramics (a) VC-Al40, (b) VC-Al35, (c) VC-Al30, (d) VC-Al25, (e) VC-Al20, and (f) VC-Al15. The
individual phases are marked by arrows.

11

B resonances of the

H. Deters, H. Eckert / Solid State Nuclear Magnetic Resonance 41 (2012) 4859

YAl3(BO3)4

YAl3(BO3)4

B2O3

Al5BO9

YBO3

YBO3

Phase X

Phase X

55

VC-Al40

VC-Al25
30

20

10

-10

30

20

10

-10

(11B) / ppm

(11B) / ppm

Fig. 9. 11B MAS NMR spectrum obtained for the vitroceramic VC-Al25 (left) and the vitroceramic VC-Al40 (right). Included are simulations based on the individual line
shape components reecting the different types of boron-containing species present.

Table 6
11
B MAS NMR analysis of the yttrium aluminoborate vitroceramics in the VC-Alx
series. nMole% values predicted from the crystallization model are given in
parentheses, see text.
Group

11

B NMR area
of individual
groups/%

11
B NMR area
of individual
phases/%n

VC-Al15

YAl3(BO3)4

B(3)-I
B(3)-II
B(3)-I
B(3)-II
B(3)
B(4)

6
17
36
14
1
26

23 (31)

B(3)-I
B(3)-II
B(3)-I
B(3)-II
B(3)
B(4)
B(3)-I
B(3)-II
B(3)-I
B(3)-II
B(3)
B(4)
B(3)-I
B(3)-II
B(3)
B(3)
B(4)
B(3)-I
B(3)-II
B(3)
B(3)
B(4)
B(3)-I
B(3)-II
B(3)
B(3)
B(4)

8
30
25
9
1
27
14
41
15
3
8
19
18
55
1
4
22
16
47
6
2
29
12
35
12
1
39

38 (44)

B2O3
Phase X
YBO3

50 (46)
1 (0)
26 (23)

signal fractional area

Phase

YAl3(BO3)4

11B

Sample

80

60

40
B2O3

YBO3

20

Al5BO9

VC-Al20(VC-Y20)
YAl3(BO3)4
B2O3

VC-Al25

Phase X
YBO3
YAl3(BO3)4
B2O3

VC-Al30

Phase X
YBO3
YAl3(BO3)4

VC-Al35

Al5BO9
Phase X
YBO3
YAl3(BO3)4

VC-Al40

Al5BO9
Phase X
YBO3
YAl3(BO3)4
Al5BO9
Phase X
YBO3

34 (33)

Phase X
15

1 (0)
27 (22)
55 (61)
18 (18)
8 (0)
19 (21)
73 (80)
1 (0)
4 (0)
22 (20)
63 (66)
6 (6)
2 (0)
29 (28)
47 (48)
12 (14)
1 (0)
39 (38)

individual line shape components as adjustable parameters. Fig. 9

shows the simulation attempts for two representative VC-Alx
samples. While the agreement is generally found quite good, the
main deviation between experimental and simulated data arises
in the region of four-coordinated boron atoms (near 3 ppm). Most
likely these discrepancies arise from the well known fact that the
spectrum of YBO3 is not well described by just a single Gaussian,
but shows a more complex lineshape with a distinct highfrequency shoulder. Table 6 lists the approximate 11B signal area

20

30
25
mole % Al2O3

35

40

Fig. 10. 11B signal fractional areas obtained for YAl3(BO3)4, YBO3, Al5BO9, glassy
B2O3, and the unknown Phase X in the yttrium aluminoborate vitroceramics of the
VC-Yx series, with x representing the Al2O3 content in mole%. Solid lines denote
the predictions from Eqs. (4) and (5), see text.

fractions, which are plotted in Fig. 10 as a function of the alumina

content. The data are compared with the theoretical area fractions
predicted from two crystallization scenarios, in both of which the
formation of the unknown crystalline phase X has been neglected
for simplicity; for this reason, the description is only approximate.
The rst scenario applies to all glasses with xo0.30. In this
scenario all of the aluminum atoms end up in crystalline
YAl3(BO3)4 (to be conrmed by 27Al NMR, see next section) and
the remaining yttrium content reacts to form crystalline YBO3,
leaving excess boron oxide behind. In this way, excess boron
oxide may be considered as a ux material whose quantity is
determined by the chemical composition, i.e. the x value. Based
on these considerations the crystallization scenario in this system
can be expressed by the following equation:
 
11501C 2x
B2 O3 0:8x Al2 O3 x Y2 O3 0:2 !
YAl3 BO3 4
3


2x
YBO3 0:62xB2 O3
0:4
4
3
According to this equation, glassy B2O3 is no longer formed for
x values equal to or greater than 0.30. In case of higher alumina
contents (x Z0.3) a second crystallization scenario takes over,
leading to crystalline YAl3(BO3)4, YBO3, and Al5BO9. Thus, all of

56

H. Deters, H. Eckert / Solid State Nuclear Magnetic Resonance 41 (2012) 4859

representative samples (VC-Al15 and VC-Al40) obtained for

sufciently long dipolar evolution times. The nondephased signal
(S) of VC-Al15 closely resembles that of glassy B2O3, suggesting
the latter to be the remaining phase in this glass ceramic. Clearly,
the B(4) units do not show any REDOR effect, further conrming
that this particular resonance is exclusively attributable to crystalline YBO3, and that there are no four-coordinated boron species
interacting with 27Al spins (no Al4B2O9). As shown earlier, the
fraction of three fold coordinated boron atoms that interact with
27
Al can be determined from these REDOR experiments. Fig. 12
(left part) illustrates the REDOR attenuation curves obtained for
all of the vitroceramic samples. The data on VC-Al20 are slightly
different from those discussed on VC-Y20 in reference [41], even
though the compositions are identical. This difference may be
attributed to the longer annealing times used in the present study
as compared to Ref. [41], potentially resulting in higher crystallinity of the present samples, but also some B2O3 loss owing to
volatilization. For VC-Al30 and VC-Al40 the REDOR attenuations
are found to be complete at longer dipolar evolution times,
indicating that all of the B(3) units are locally in contact
with 27Al. This is basically consistent with the proposed crystallization scenario (Eq. (5)) according to which all of the threecoordinated boron units arise from to YAl3(BO3)4 and Al5BO9,
respectively. In case of the samples with x o0.30, the experimentally found fractions of three-coordinated boron species arising
from YAl3(BO3)4, which are expected to show a REDOR effect, are
generally a little higher than those predicted from Eq. (4). For
example, for the sample with x 0.15 Eq. (4) predicts that 40% of

the aluminum atoms end up in crystalline YAl3(BO3)4 and Al5BO9

(to be conrmed by 27Al NMR, see next section), while the
remaining yttrium atoms end up in crystalline YBO3 (to be probed
by 89Y NMR, see further below), leaving no additional phases
behind. This crystallization scenario is expressed by the following
equation:
B2 O3 0:8x Al2 O3 x Y2 O3 0:2
11501C

!0:5xYAl3 BO3 4 x0:1YBO3 x0:3Al5 BO9

Fig. 10 indicates good agreement of the experimental data

with the two proposed crystallization scenarios in their appropriate compositional regions. Only some small deviations can be
noticed, which are most likely connected to the presence of the
additional phase X and to the fact that the stoichiometries of the
individual crystalline compounds may not be perfectly correct.
Nevertheless, it seems that the quantities of the major crystallization products can be more or less directly predicted from the
glass composition. Still, it can be argued that the limited resolution observed in the 11B MAS NMR spectra might generate some
experimental uncertainty concerning the quantitative contributions of the individual phases. Thus, independent verication by
additional experiments is necessary.
As shown in reference [41], 11B{27Al} REDOR experiments are
useful for spectroscopically separating the fractions of boron
atoms that do or do not interact with 27Al. Fig. 11 illustrates the
REDOR difference signals (S0S), the non-dephased signals (S),
and the regular rotor synchronized spin echoes (S0) of two

VC-Al40
NTR = 0.0093 s

VC-Al15
NTR = 0.0093 s

S0-S
S0-S

S0 and S

20

30

10

-10

S0 and S

30

20

10

-10

(11B) / ppm

(11B) / ppm

Fig. 11. Spectral editing via 11B{27Al} REDOR, demonstrated for the vitroceramics VC-Al15 (left) and VC-Al40 (right) for a dipolar evolution time of 0.0093 s. In case of VCAl15 the nondephased signal (S) emphasizes the contributions from glassy B2O3 and crystalline YBO3.

VC-Al40
1.0

VC-Al30

1.0

VC-Al25

0.8

0.8

0.6
VC-Al15
0.4

S/S0

S/S0

VC-Al20

VC-Al15

0.6
0.4

YAl3(BO3)4*0.40
0.2

0.2

0.0

0.0
0.00

0.01
NTR / s

0.02

0.03

0.00

0.01
NTR / s

0.02

0.03

Fig. 12. Left: 11B{27Al} REDOR curves measured for the B(3) units of the yttrium aluminoborate vitroceramics VC-Al15, VC-Al20, VC-Al25, VC-Al30, and VC-Al40. Right:
comparison of the data obtained from VC-Al15 with the predicted REDOR curve according to the crystallization model of Eq. (4).

H. Deters, H. Eckert / Solid State Nuclear Magnetic Resonance 41 (2012) 4859

the B(3) units belong to YAl3(BO3)4 and 60% to glassy B2O3, which
do not interact with 27Al. Experimentally, however, the normalized difference signal approaches 0.47 instead of 0.40 (see right
part of Fig. 12). This deviation may arise either from volatilization
losses or may also be related to the formation of the Phase X.
Thus, it may be concluded that the unknown phase is formed at
the expense of the formation of glassy boron oxide and, further,
that 11B in this particular phase is expected to show heteronuclear dipolar couplings with 27Al nuclei. In line with this
interpretation, it is worth noting that up to 8% of the boron atoms
(about 10% of the three-coordinated boron units) are attributed to
the Phase X with regard to the tting results obtained before.
3.2.3. 27Al MAS and TQMAS NMR
The left part of Fig. 13 illustrates the glass-to-ceramic transition
for a representative sample, indicating that devitrication is accompanied by dramatic structural changes in the aluminum environment.
The 27Al MAS NMR spectra of the vitroceramics studied in this section
are shown in the right part of Fig. 13 along with the spectrum of
crystalline Al5BO9. For the vitroceramics VC-Al20 and VC-Al25 (with
xo0.3), only one resonance located at about  1.0 ppm can be seen
in the spectra. In principle, this sharp line can be easily attributed to
crystalline YAl3(BO3)4, in excellent agreement with the crystallization
model according to Eq. (4). No additional signals attributable to the
unknown Phase X can be identied in these spectra so that the issue
of Phase X cannot be addressed here. Possibly the 27Al signal of this

57

phase cannot be resolved in the spectrum, owing to its low

concentrations and/or line-broadening effects arising from strong
quadrupolar interactions. At high alumina contents (xZ0.3), the
spectra of the corresponding vitroceramics reveal the formation of
crystalline Al5BO9. TQ MAS NMR experiments are useful for creating a
signicantly enhanced resolution, which ensures that all resonances
found in the spectra are unambiguously due to crystalline YAl3(BO3)4
and Al5BO9 and excludes the presence of any additional resonances
hidden underneath the dominant signals. No such extra signals are
visible in the two-dimensional spectrum obtained for a representative
sample (data not shown) and only the individual aluminum environments associated with crystalline YAl3(BO3)4 and Al5BO9 can be
identied. Thus, the 27Al MAS NMR spectra of the vitroceramics can
be simulated using the tting parameters obtained for both crystalline compounds, leaving only the amplitudes of the individual line
shape components as adjustable parameters. Such a simulation
attempt is schematically shown in the left part of Fig. 14. The results
of the simulations are summarized in Table 7 and illustrated in the
right part of Fig. 14 as a function of alumina content. The signal
fractions predicted from Eqs. (4) and (5), respectively, are included in
the same gure, showing excellent agreement with the proposed
crystallization models.
3.2.4. 89Y NMR
Finally, it is worth noting that the crystallization model
according to Eqs. (4) and (5) is also quantitatively consistent with

Al5BO9
VC-Al40
VC-Al35

Al40 (glass)

VC-Al30
VC-Al25
VC-Al20
VC-Al40
VC-Al15
100

50

-50

150

100

(27Al) / ppm

50

-50

-100

-150

(27Al) / ppm

Fig. 13. Left: glass-to-vitroceramic transitions observed for the sample with x 0.4 (40 mol% of Al2O3) by
aluminoborate vitroceramics in the VC-Alx series and of crystalline Al5BO9.

27

Al MAS NMR. Right:

27

Al MAS NMR spectra of yttrium

27Al

fractional signal area

100
YAl3(BO3)4

80
60
40

Al5BO9

20

VC-Al40
0
100

50

(27Al) / ppm

-50

15

20

25
30
mole % Al2O3

35

40

Fig. 14. Left: 27Al MAS NMR spectrum obtained for the ceramic VC-Al40. Included are simulations based on the individual line shape components reecting the different
aluminum environments in YAl3(BO3)4 and Al5BO9. Right: fractional signal areas of YAl3(BO3)4 and Al5BO9 in the yttrium aluminoborate ceramics in the VC-Alx series, with
x representing the Al2O3 content in mole%. Solid lines illustrate the predictions from Eq. (4) (x o 30 mol%) and Eq. (5) (xZ 30 mol%), respectively.

58

H. Deters, H. Eckert / Solid State Nuclear Magnetic Resonance 41 (2012) 4859

Table 7
27
Al MAS NMR analysis of the ceramics in the VC-Alx series. nFractional areas predicted from the crystallization model are given in
parentheses.
Sample

Phase

Group

27
Al NMR area
of groups (72%)/%

27
Al NMR area
of phases (7 2%)/%n

VC-Al15
VC-Al20
VC-Al25
VC-Al30

YAl3(BO3)4
YAl3(BO3)4
YAl3(BO3)4
YAl3(BO3)4
Al5BO9

YAl3(BO3)4
Al5BO9

VC-Al40

YAl3(BO3)4
Al5BO9

100
100
100
89
2
2
3
4
61
8
6
9
16
36
14
8
18
24

100 (100)
100 (100)
100 (100)
89 (100)
11 (0)

VC-Al35

Al(6)
Al(6)
Al(6)
Al(6)
Al(4)
Al(5)-I
Al(5)-II
Al(6)
Al(6)
Al(4)
Al(5)-I
Al(5)-II
Al(6)
Al(6)
Al(4)
Al(5)-I
Al(5)-II
Al(6)

61 (63)
39 (37)

36 (36)
64 (64)

YBO3
YAl3(BO3)4
VC-Al40Yb0.5

Al30Yb0.5
(glass)

VC-Al30Yb0.5

VC-Al30Yb0.5
VC-Al20Nd0.75

500

400

300

200

100

(89Y)

-100 -200 -300 400

300

100

200
(

89Y)

/ ppm

Fig. 15. Left: glass-to-vitroceramic transitions observed for a Yb-doped sample with x 0.3 (30 mol% of Al2O3) by
different yttrium aluminoborate vitroceramics in the VC-Alx series.

the 89Y MAS NMR results. Fig. 15 summarizes the 89Y MAS-NMR
spectra of the glass-to-vitroceramic transitions for three representative compositions doped with Yb3 or Nd3 to increase
spin-lattice relaxation rates. The YBO3 phase is observed as a
broad peak at 117 ppm (the two distinct lines at 117.6 and
113.8 ppm observed in pure YBO3 are not resolvable here
because the paramagnetic broadening is too strong), while the
YAl3(BO3)4 phase is observed at 135 ppm. The experimental ratios
of 34:66, 45:55, and 23:77 are in good agreement with the ratios
of 33:67, 50:50, and 25:75 predicted from the appropriate
Eqs. (4) or (5), for the respective samples investigated. To ensure
quantitative conditions, a sufciently long relaxation delay
(5000s) was used in this case. The absence of any other
resonances suggests that essentially the entire yttrium content
of the glass ends up in crystalline YBO3 and YAl3(BO3)4. In this
way, the 89Y NMR data give no evidence for the formation of
Y8.67B2(BO3)2O13. On the other hand, the signal intensity in this
compound is expected to be spread over nine resonances corresponding to the crystallographically distinct yttrium sites; this
could make the detection of low levels of this compound rather
difcult.

89

-100

/ ppm

Y MAS-NMR. Right:

89

Y MAS-NMR spectra of three

4. Conclusions
In summary, the crystallization of yttrium aluminoborate
glasses along the composition line (B2O3)0.8  x(Al2O3)x(Y2O3)0.2
has been investigated by means of X-ray powder diffraction and
various solid-state NMR techniques. On the basis of 11B, 27Al, 89Y
NMR measurements as well as 11B {27Al} REDOR experiments, a
detailed quantitative description of the crystallization product
distribution as a function of composition could be presented. In
the system investigated, two main crystallization scenarios need
to be considered. For small alumina contents (x o0.3) the major
products of the phase segregation processes are crystalline YBO3,
crystalline YAl3(BO3)4, and residual glassy B2O3, whereas the
latter compound is no longer present in case of ceramics with
high alumina contents (xZ0.3). Instead, crystalline Al5BO9 is
formed. As a consequence it seems that the entire yttrium
inventory of the glassy precursor is distributed into the crystallization products YBO3 and YAl3(BO3)4, and no additional yttriumcontaining phase is formed. These ndings are basically identical
to those obtained for the VC-Yx series studied previously [41].
Nevertheless, 11B TQ MAS and 11B{27Al} REDOR data could reveal

H. Deters, H. Eckert / Solid State Nuclear Magnetic Resonance 41 (2012) 4859

the formation of an additional crystalline phase in which boron is

expected to be coordinated by three oxygen atoms and closely
interacting with aluminum. The nature of this extra phase is
subject of further investigations in our laboratory.

Acknowledgments
H.D. thanks the NRW Graduate School of Chemistry for a
personal fellowship. We thank Dr. Jinjun Ren for conducting the
electric eld gradient calculations using the WIEN2k code, and
Professor Andrea Simone Stucchi de Camargo for fruitful
discussions.
References
[1] D. Jaque, J.C. Lagomacini, C. Jacinto, T. Catunda, Appl. Phys. Lett. 89 (2006)
121101.
[2] S. Jiang, M. Myers, N. Peyghambarian, J. Non-Cryst. Solids 239 (1998) 143.
[3] A.S.S. De Camargo, C. Jacinto, T. Catunda, L.A.O. Nunes, Appl. Phys. B 83 (2006)
565.

[4] C. Honninger,
R. Paschotta, M. Graf, Appl. Phys. Lett. B:Lasers Opt. 69 (1999)
3.
[5] A.S.S. De Camargo, I.A.A. Terra, L.A.O. Nunes, M.S. Li, J. Phys. Cond. Matter 20
(2008) 255240.
[6] C.N. Santos, D. Mohr, W.F. Silva, A.S.S. De Camargo, H. Eckert, M.S. Li,
M.V.D. Vermelho, A.C. Hernandes, A. Ibanez, C. Jacinto, J. Appl. Phys. 106
(2009) 023512.
[7] I.N. Chakraborty, H.L. Rutz, D.E. Day, J. Non-Cryst. Solids 84 (1986) 86.
[8] H.L. Rutz, D.E. Day, C.F. Spencer, J. Am. Ceram. Soc. 73 (1990) 1788.
[9] J. Rocherulle, J. Mat, Sci. Lett. 22 (2003) 1127.
[10] N. Hemono, J. Rocherulle, M. Le Floch, B. Bureau, P. Benard-Rocherulle, J.
Mater. Sci. 41 (2006) 445.
[11] R. Brow, D.R. Tallant, G.L. Turner, J. Am. Ceram. Soc. 80 (1997) 1239.

[12] J.C.C. Chan, R. Ohnesorge, K. Meise-Gresch, H. Eckert, W. Holand,

Chem.
Mater. 13 (2001) 4198.

[13] W. Holand,
C. Ritzberger, E. Apel, V. Rheinberger, R. Nesper, H. Krumeich,

C. Monster,
H. Eckert, J. Mater. Chem. 18 (2008) 1318.
[14] B. Chen, U. Werner-Zwanziger, M.L.F. Nascimento, L. Ghussn, E.D. Zanotto,
J.W. Zwanziger, J. Phys. Chem. C 113 (2009) 20725.

59

[15] J.G. Longstaffe, U. Werner-Zwanziger, J.F. Schneider, M.L.F. Nascimento,

E.D. Zanotto, J.W. Zwanziger, J. Phys. Chem. C 112 (2008) 6151.
[16] A. Rawal, X. Wei, M. Akinc, K. Schmidt-Rohr, Chem. Mater. 20 (2008) 2583.
[17] L. Van Wuellen, T.K.-J. Koester, Solid State Ionics 180 (2009) 141.

[18] M. Bertmer, L. Zuchner,

J.C.C. Chan, H. Eckert, J. Phys. Chem. B 104 (2000)
6541.
[19] J.C.C. Chan, M. Bertmer, H. Eckert, J. Am. Chem. Soc. 121 (1999) 5238.
[20] H. Deters, A.S.S. De Camargo, C.N. Santos, C.R. Ferrari, A.C. Hernandes,
A. Ibanez, M.T. Rinke, H. Eckert, J. Phys. Chem. C 113 (2009) 16216.
[21] M. Garsche, E. Tillmanns, H. Almen, H. Schneider, V. Kupcik, Eur. J. Mineral 3
(1991) 793.
[22] A. Medek, J.S. Harwood, L. Frydman, J. Am. Chem. Soc. 117 (1995) 12779.
[23] J.P. Amoureux, C. Fernandez, S. Steuernagel, J. Magn. Reson. A 123 (1996) 116.

[24] L. Zuchner,
J.C.C. Chan, W. Muller-Warmuth,
H. Eckert, J. Phys. Chem. B 102
(1998) 4495.
[25] D. Massiot, F. Fayon, M. Capron, I. King, S.L. Calve, B. Alonso, J.O. Durand,
B. Bujoli, Z. Gan, G. Hoatson, Magn. Reson. Chem. 40 (2002) 70.
[26] A. Medek, L. Frydman, J. Braz. Chem. Soc. 10 (1999) 263.
[27] T. Gullion, J. Schaefer, J. Magn. Reson. 81 (1989) 196.
[28] T. Gullion, Magn. Reson. Rev. 17 (1997) 83.
[29] S. Steuernagel, Private Communication, Bruker BioSpin, Rheinstetten, 2008.
[30] L.N. Swink, G.B. Carpenter, Acta Cryst. 22 (1967) 602.

[31] G. Kunath, P. Losso, S. Steuernagel, H. Schneider, C. Jager,

Solid State Nucl.
Magn. Reson. 1 (1992) 261.

[32] D. Massiot, D. Muller,

Th. Hubert,
M. Schneider, A.P.M. Kentgens, B. Cote,
J.P. Coutures, W. Gessner, Solid State Nucl. Magn. Reson. 5 (1995) 175.

[33] I. Hung, A.P. Howes, P. Anupold,

A. Samoson, D. Massiot, M.E. Smith,
S.P. Brown, R. Dupree, Chem. Phys. Lett. 432 (2006) 152.
[34] K.J.D. MacKenzie, M.E. Smith, T.F. Kemp, D. Voll, Appl. Magn. Reson. 32 (2007)
647.
[35] Z. Gan, P. Gorkov, T.A. Cross, A. Samoson, D. Massiot, J. Am. Chem. Soc. 124
(2002) 5634.
[36] H.P. You, G.Y. Hong, X.Q. Zeng, Chem. Res. Chin. Univ. 16 (2000) 301.
[37] K.J. Griesser, A. Beran, D. Voll, H. Schneider, Miner. Pet. 92 (2008) 309.
[38] M. Fisch, T. Armbruster, D. Rentsch, E. Libowitzky, T. Pettke, J. Solid State
Chem. 184 (2011) 70.
[39] J.C.C. Chan, Hellmut Eckert, J. Magn. Reson. 147 (2000) 170.
[40] J.D. Epping, W. Strojek, H. Eckert, Phys. Chem. Chem. Phys. 7 (2005) 2384.
[41] H. Deters, A.S.S. De Camargo, C.N. Santos, H. Eckert, J. Phys. Chem. C 114
(2010) 14618.
[42] R.X. Fischer, V. Kahlenberg, D. Voll, K.J.D. Mackenzie, M.E. Smith, B. Schnetger,
H.J. Brumsack, H. Schneider, Am. Miner. 93 (2008) 918.
[43] A. Vegas, F.H. Cano, S. Garcia-Blanco, Acta Cryst. B 33 (1977) 3607.
[44] J.H. Lin, S. Zhou, L.Q. Yang, G.Q. Yao, M.Z. Su, L.P. You, J. Solid State Chem. 134
(1997) 158.